Linkable administrative files: Family information and existing data

Abstract

Linkable administrative data have facilitated research incorporating files from various government departments. Examples from Canada, Australia, and the United Kingdom highlight the possibilities for improving such work. After expanding on comparisons of linkable administrative data with several famous studies, we forward suggestions on improving research design and expanding use of family data. Certain characteristics of administrative data: large numbers of cases, many variables for each individual, and information on parents and their children, provide building blocks for implementing these proposals. Traditional longitudinal epidemiological approaches can be modified to facilitate a quasi-experimental perspective. Incorporating multiple research designs within the same project handles threats to validity more easily. Family data provide a number of opportunities for both same-generation and intergenerational research. Risk factors associated with a number of conditions can be studied. Bad events can affect all family members, and cross-sectoral information can extend analyses beyond health to include educational outcomes. Parent/child linkages suggest several lines of research exploring within-family relationships. Complicated data call for family identification systems to estimate project practicality. Manitoba administrative data are presented to illustrate one such system. Problems in maintaining core data element –­ such as marital status ­­­– have been highlighted. The productivity and potential of cross-sectional, longitudinal, and life course research using existing information have emphasised the value of investments in such work.